Draft (June 25, 2002) for A. Mele and P. Rawlings, eds., Rationality (New York: Oxford University Press, forthcoming)


Patricia Greenspan

University of Maryland,

College Park

The category of emotions covers a disputed territory, but clear examples include fear, anger, joy, pride, sadness, disgust, shame, contempt and the like. Such states are commonly thought of as antithetical to reason, disorienting and distorting practical thought. However, there is also a sense in which emotions are factors in practical reasoning, understood broadly as reasoning that issues in action. At the very least emotions can function as "enabling" causes of rational decision-making (despite the many cases in which they are disabling) insofar as they direct attention toward certain objects of thought and away from others. They serve to heighten memory and to limit the set of salient practical options to a manageable set, suitable for "quick-and-dirty" decision-making.

Current research in neuroscience and other areas indicates that practical reasoning in this sense presupposes normal emotional development and functioning (see, e.g. Damasio 1994). Evolutionary accounts of emotion (e.g., Cosmides and Tooby 1992) stress their role in rational design of the human organism. Contemporary philosophy of emotion attempts something stronger, however, in according emotions a role in practical reasoning. Making this an integral role--understanding emotions as functioning within practical reasoning rather than just as spurs to it--means interpreting emotions in normative terms, as providing or expressing potential reasons for action, and as themselves subject to rational assessment and control, contrary to the traditional view of emotions as "passive" phenomena.

1. Emotions as Evaluative

The dominant approach in contemporary philosophy rests on assigning emotions an evaluative content (see, e.g., Bedford 1957, Solomon 1976, Lyons 1980, Budd 1985, Davis 1987, de Sousa 1987, Roberts 1988, Greenspan 1988, Nussbaum 1993, Stocker 1996). Human emotions typically are directed toward "intentional objects" in the sense of being about something, real or imagined but in any case an object of thought. Emotions that represent their objects in some positive or negative light (as most do) may be said to have a content expressible by an evaluative proposition. Where the evaluation either is or implies an evaluation of some future contingency that the agent (the one who undergoes the emotion) can bring about or avoid, the emotion counts as a reason for or against action.

The rational bearing of emotions on action in these terms is captured roughly by the first or major premise of an Aristotelian practical syllogism, Aristotle's three-line schema of practical reasoning (see 1925: Book VII). In contemporary terms the major premise expresses a pro- or con-attitude toward something that the second or minor premise tells one how to attain or avoid. It evaluates something as good or bad, desirable or worth avoiding--or simply as an object of a current desire or aversion.

For instance, fear may be said to represent its object (what one is fearful of or about; a certain event, or a likely cause of it) as a threat, a possible harm. Anger represents its object (another person, or his performance of some action) as already a cause of harm or offense that now calls for retaliation. Despite differences between the propositional content of these two emotions--one describes a possibility, the other a fact (Gordon 1987)--both include a reason for avoiding some future contingency: the presence or occurrence of what is feared or a failure to retaliate for the cause of anger.

Emotions also involve a corresponding affective element, derived from pleasure or pain on Aristotle's account of emotions (see 1941), that amounts to a good or bad state of the agent and hence supplies a reinforcing reason for action (Greenspan 1988). In contemporary decision-theoretical terms, the affective element modifies the "payoff structure" of the situation: the array of potential costs and benefits of alternative responses open to the agent. The discomfort experienced in fear or anger, for instance, provides the agent with a further (pro tanto) reason for acting to change the situation that provokes discomfort. On the other hand, joy or pride, as positive evaluations of some state of affairs or of oneself, do not provide a reason to change anything, but their affective aspect yields a further reason to sustain the conditions that make the evaluation appropriate. However, the focus of the agent's attention in all these cases is normally the evaluative content of emotion, not her own state of feeling.

Apart from its role as a source of reinforcing reasons, the affective or feeling aspect of emotion is useful to practical reasoning just insofar as it serves to hold in mind the evaluative content of emotion without explicit reflection. For instance, while driving on the highway one does not have to deliberate at length about possible bad consequences of swerving and their relevance to the task of steering straight. To the extent that steering straight is not just automatic, and the driver wanders out of lane, an anxious awareness of the possibility of an accident brings her quickly back to the task.

This way of anticipating practical eventualities in everyday life corresponds roughly to neuroanatomist Antonio Damasio's (1994) understanding of emotions as "somatic markers." Damasio's view is put forth to explain why cases of emotional impairment due to brain lesion (such as the famous nineteenth-century case of Phineas Gage) involve a loss of practical reasoning ability. Emotions serve to "mark" practically significant thoughts with bodily (and hence affective) indicators of past experience. On an evaluative account, characteristic thoughts have come to be contents of emotion--and part of what identifies them as the types of emotion they are: fear, anger, joy, pride, and so forth.

A question for philosophers who accept an evaluative account is whether and in what sense emotions themselves are subject to rational assessment. It is only in a qualified sense that an irrational evaluative stance could be said to justify action. Hume's famous denial (see 1978) that the passions can be rationally assessed at all was based on his interpretation of them as nonrepresentational. But contemporary evaluative accounts interpret emotions, in effect, as representing evaluative propositions: that something portends harm or deserves retaliation or is a stroke of good luck or praiseworthy.

2. Belief-based Views

The evaluative implications of emotion have been characterized thus far without commitment to a particular contender for capturing their essential nature. This is the philosopher's usual first question, and it was set up as the focus of contemporary debate on emotion by James (1884). However, as Rorty (1980) and others have noted (cf. Griffiths 1997), emotions do not form a natural class. A later approach (de Sousa 1987, Greenspan 1988) begins with questions of rationality, treating cases of rational emotional response as paradigmatic. Emotions in lower animals or infants (along with some elements of our own experience that derive from them unchanged) can be understood from this standpoint as deficient instances of full-blown emotions functioning normally in adult human life.

As applied to accounts of the nature of emotions, the evaluative approach is often discussed under the heading of "cognitivism," which interprets emotions themselves as amounting to or containing cognitions, usually seen as mental states representing evaluative propositions. The most straightforward version of the cognitivist view is "judgmentalism," which understands the nature of emotions in terms of judgments: assertive propositional attitudes, assessable by the ordinary evidential standards applied to beliefs. The view traces back to the Stoics (see Nussbaum 1993), though the Stoics interpreted our ordinary emotions as "confused" rather than rational judgments.

The contemporary version of judgmentalism was introduced into the Anglo-American philosophical literature by Bedford (1957), in a post-Wittgensteinian response to the Cartesian tradition of identifying emotions with particular passions or feelings, as in Hume (cf. Descartes 1970, James 1884). The view was extended by Kenny (1963) and Pitcher (1965), and it was brought into connection with other philosophical traditions and with work in psychology and other areas by Solomon (1976), Neu (1977), Lyons (1980), and Ben-Ze'ev (2000). Davidson (1980) defends a version of judgmentalism for pride and related emotions.

Proponents of judgmentalist accounts do not always seem to have the same thing in mind by "judgment." Diverse cases indicate, moreover, that the view needs some sort of qualification to capture situations where emotions register evaluative thoughts that may not rest on evidence supporting judgments in the sense of all-things-considered beliefs but nonetheless seem to be rational to hold in mind and to act in light of.

Some irrational cases such as phobias (fear of a dog one knows to be harmless, say) that undermine judgmentalism as a view of the nature of emotions are explored at length in Pitcher (1965), Stocker (1987), and Greenspan (1981) and (1988). However, there are also many morally or motivationally important emotions that one might be inclined to assess as rational but that turn on imagining some state of affairs that is not applicable (at least in current terms) to oneself.

Cases of empathetic emotion, for instance, seem to involve putting oneself in the position of another person, typically someone suffering harm. However, acting effectively out of sorrow felt on behalf of someone else often depends on not losing sight of the fact that one is not really in the same position oneself. Similarly for anticipatory emotions such as guilt at the thought of something one might (or might not) do. There are also "emotional self-management" cases--of generating upbeat feelings, say, by "looking at the bright side" of some more problematic object of evaluation, or looking past someone's flaws to sustain love--that are crucial to mental health and to thriving interpersonal relationships.

3. Modifying Judgmentalism for Rationality Issues

One might distinguish a possible variant of judgmentalism meant to apply specifically to rationality issues. Whatever their nature as psychological states, on this account, emotions can be understood in terms of some sort of claim they make about the agent's situation. Whether a given emotion is rationally appropriate depends on whether this corresponding judgment would be warranted as a belief--whether or not the emotion actually involves that belief.

Emotions that do not conform to this model such as empathetic or anticipatory emotions might still sometimes be said to play a limited instrumental role in practical reasoning as cases in which an irrational mental state serves as part of an overall rational strategy. This idea of emotional motivation as "rational irrationality" comes up particularly in discussion of evolutionary "design strategies" that might account for genetically programmed emotional proclivities underlying human morality (see Frank 1988).

Thus, one could grant that feeling sympathetic sorrow over someone else's loss would not entail a belief that one has suffered some sort of loss oneself, or warrant for such a belief. But on the present suggestion that is just because it is not really rational in the sense of being appropriate to one's own situation. The feeling may still be important to sustaining mutually supportive personal relationships, and we can encourage it in ourselves with that end in mind, though to preserve its other-directed quality, we usually try to keep such calculations subliminal, or refer them to evolutionary design.

Whether this is a satisfying answer for all cases depends in part on how we think of rationality. In common parlance the term is sometimes used in a "thick" sense that would be implausible for most emotions insofar as it connotes explicit calculation; by contrast, Gibbard 1991 uses the term in an extremely "thin" sense in which it just conveys general positive endorsement. But there seems to be room between these poles for a use that captures distinctions we commonly make between reasonable and unreasonable emotional reactions, appropriate and inappropriate in what seems to be roughly a rational sense, having to do with some notion of fit to the circumstances that constitute grounds or evidence for emotion. We might think of this medium-thick sense as "representational" rationality insofar as it depends on assigning emotions intentional content.

However, our ordinary notion of emotional appropriateness seems to allow for variable standards of evidence and for more than one option in many cases. For instance, under ordinary circumstances it might be reasonable enough to feel somewhat worried about how a recent medical test will turn out, even in the absence of grounds for real agitation. At the same time, there may be no rational barrier to foregoing any thought about the matter or even feeling hopeful or confident. Being seized with fear would be irrational, let us say, not just because it would do no good, but more fundamentally because there are no grounds for holding in mind with that intensity any subset of the available evidence that would warrant the belief that one is in danger.

An emotion might come out as rationally appropriate, on this way of looking at things, even in cases where the corresponding all-things-considered belief would not be warranted. In other words, emotional appropriateness does not seem to be determined by a unitary overall assessment of the available evidence, of the sort presupposed by belief warrant. There is at most one adequately warranted belief (over the threshold for assent) about the probability of a bad outcome, relative to the evidence available to the agent. Belief may admit of degrees, but where the evidence is slight, it would be irrational to adopt even a low-grade belief, if that entails adopting conflicting beliefs. However, the emotional analogue (ambivalent feelings, in some cases even intense feelings) would seem to be rationally acceptable--rational in a relatively undemanding sense that allows for options (Greenspan 1980).

We can think of this as "perspectival" rationality. Moderate worry and hope in the case just outlined may each be rational from different perspectives--with reference to different subsets of the available evidence, each worthy of the relevant degree of emotional attention though not requiring it. A perspectival account of emotional rationality would appeal to a notion of warrant for a prima facie belief, as determined from some practically significant standpoint of evaluation (Greenspan 1988, 1995). So it still represents a fairly conservative move away from judgmentalism on rationality issues--though it does move beyond the view just distinguished, which makes out emotional appropriateness as dependent on warrant for a corresponding belief. The perspectival account associates emotions with evaluative thoughts rather than strictly judgments in the sense of beliefs (or acts of forming beliefs), but it still allows that the rationality of emotions can be explained in terms of belief warrant.

4. A Causal/Historical Approach

In a more radical departure from judgmentalism, emotions might be thought to be justified rationally by reference to their causal histories rather than some notion of evidence, of the sort that applies to beliefs. In light of the prominence of causal theories in other areas of philosophy, a causal/historical approach to emotions would seem to provide an alternative to judgmentalism and its variants. Some approach along these lines is suggested by recent evolutionary treatments of emotion in philosophy (e.g., Gibbard 1991, Griffiths 1997). However, the only example of it that has been formulated to apply to issues of rationality in the sense indicated is the "paradigm scenarios" view in de Sousa (1987), which is still phrased in roughly cognitivist terms.

On de Sousa's account, whose initial sources are psychoanalytical, emotions are based on infantile response-tendencies such as smiling or crying that become intentional or object-directed, and thereby come to be signs or elements of emotions, by association with a typical story about what prompts the response. The story is what de Sousa calls a "paradigm scenario," meaning that it sets the standard for a particular emotion type, essentially by establishing criteria of appropriateness. Originally, what is involved here is the child's interaction with caretakers, which serves to modify natural responses by a form of dramatic role-playing (eliciting or exchanging smiles, say, or conveying empathy). Later cultural influences can modify the emotion, though it will remain appropriate in some form as a response to the paradigm scenario.

A central example de Sousa wants to defend as appropriate in these terms, in response to some feminist arguments, is jealousy, or jealous rage, with psychoanalytic sources shaped by monogamous culture. On his account the emotion is set up as a response to a situation of "being robbed by another of vital physical attention" (185). That is how it gets its status and meaning as an intentional response, so it necessarily counts as appropriate on that basis.

It is unclear, though, whether the paradigm scenarios account can explain more than the appropriateness of (at least some central cases of) a given type of emotion. De Sousa's essential point about rationality in the sense of appropriateness seems to be just that there cannot be an emotion type of which all possible instances are inappropriate. However, it does not follow from this that a given emotion instance should be assessed as appropriate on the basis of its resemblance to the paradigm scenario for its type. More specific rationality questions--under what circumstances jealousy is appropriate in adult life, say--remain open.

De Sousa speaks of the rationality of emotions in several senses, but overall he seems to be most concerned with rationality in the sense of objectivity. On that understanding emotions are rational in general terms to the extent that they represent real properties of the world, existing independently of emotions. This amounts to an emotion-based version of the metaethical question of the objectivity of values, since the properties in question, corresponding to different types of emotion, are "axiological" or value properties.

However, recent views within the metaethical literature would seem to allow for emotional rationality in a sense independent of objectivism about values (see, e.g., Gibbard 1990, Blackburn 1998; cf. Greenspan 1995 and 1998, D'Arms and Jacobson 2000). Objectivity need not be our concern when we speak of the rationality of particular emotions in the sense of appropriateness but rather their fit to the reasons available to the agent, leaving it as a further question whether reasons or values in general can be thought of as objective.

De Sousa at some points limits himself to a claim of "minimal rationality" or basic intelligibility of emotion types, but he also applies the paradigm scenarios view to the question of the correctness or incorrectness of particular emotion instances, which his discussion makes out as dependent on the notion of normality (cf., e.g., 187, 201). But some emotional responses that are normal and understandable might be inappropriate. We can make sense of a claim, for instance, that the paradigm for male jealousy--what degree and sort of attention the jealous agent feels he is entitled to and hence can be "robbed" of--has been distorted by male upbringing in a certain culture. So even the received paradigm of an emotion can be skewed or off the mark.

Applying the paradigm scenarios approach to cases would seem to involve appeal to an interpretation of the situation that might be summed up in a proposition: that one has been robbed of vital physical attention, in the jealousy example. How this gets extended or modified by culture (extended to other forms of attention, perhaps; or modified by challenging the criteria of entitlement to attention) may be a matter of historical derivation, but the criteria at a given time and relative to a given culture would seem to be specifiable in terms that fit the belief-based approach outlined earlier. De Sousa rejects the analogy between emotions and beliefs as a basis for his approach to rationality issues at least partly because he understands rational beliefs as arrived at in a certain way, on the basis of reasoning (see 5; cf. 197f.). However, this is not part of the belief-based approach as here intended.

It is not clear how the paradigm scenarios view could handle cases without appeal to belief-based criteria of rationality. For example, imagine someone who momentarily feels jealous anger when his wife exchanges glances with another male at a party. In de Sousa's terms the man is reacting to what he sees as a robbery of vital attention--in terms of Aristotle's definition of anger (see 1941: 1378a31-b5), an unjustified slight, or at any rate an indication that a slight is imminent. But this is an evaluative judgment interpreting past events and their natural and conventional meanings: what a glance means or can mean, what legitimate expectations a relationship confers, that a glance involves or might lead to intimacies that violate those expectations, and that violating them inflicts loss and insult on the agent.

The emotional reaction in such a case might sometimes be assessed as inappropriate to the current situation--if the agent really knows, say, that his wife and the recipient of her glance, a colleague in her area, are merely reacting to a professional faux pas on the part of someone else at the party--even where situational cues do naturally give rise to jealousy because of their superficial resemblance to the paradigm scenario. To explain which cues render an instance of emotion appropriate, rather than merely natural or normal or understandable, we seem to need at least implicit reference to the notion of a propositional content, as what the emotion still essentially "claims" about the situation, even if the reasons for it have changed since the paradigm scenario was established.

5. Emotional Strategies

A variant of the general causal-historical approach that rejects reference to propositional attitudes is the evolutionary account in Griffiths (1997). Though Griffiths is mainly concerned with the question of the nature of emotions rather than rationality issues, his argument turns out to rest at crucial points on a familiar way of setting up emotions in opposition to reason. He continually adverts to a definition of emotions as "irruptive" or passive states, interfering with the agent's stable goals, and not themselves subject to rational strategy (see, e.g., 155-57, 233f., 242ff.; cf. 9, 16, 118, 120). Those cases in which emotions appear to be used strategically cannot be genuine emotions on his account.

Griffiths particularly has in mind emotions based on the social pretense that we lack rational control over a certain pattern of behavior. He draws on anthropological examples from defenders of "social constructivist" views of emotions (see, e.g., Averill 1980, Lutz 1986) in which the conditions of particular syndromes of emotions seem to be based on social conventions. A parallel from contemporary urban life is the "Rambo" syndrome of violent male rage; some conceptions of romantic love might also qualify. However, Griffiths' point is that whatever is felt in such cases would be disqualified from inclusion in a class of genuine emotions.

However, it seems, on the contrary, that we can make sense of a self-fulfilling social pretense of emotion in the example of male jealousy drawn from de Sousa. Imagine someone working himself into states of jealous anger on flimsy or imagined grounds (or even on good grounds that he normally would ignore) in order to provoke a certain kind of interaction with his spouse--to exert control, perhaps, or perhaps just to bring about an occasion to express and enhance affection.

An attempt to understand emotions as voluntary actions, things we do rather than states that come over us, was part of Solomon's (1973) version of judgmentalism, based on Sartre's (1962) conception of emotions as "magical transformations of reality" but meant to allow for rationality. However, we need a less extreme version of the view to modify the usual picture of rationality and emotions. The rational role of emotions depends on the fact that our control over them is limited.

Thus, for instance, Frank (1988) assigns emotions a crucial role in answer to what he calls "the commitment problem," the problem of how we can demonstrate to others effectively our commitment to future actions against narrow self-interest, as needed to secure their willingness to take part in cooperative schemes (on the model of the prisoner's dilemma, where each agent has to resist inducements to confess a joint crime). Emotions are as useful as they are to social communication just by virtue of the fact that they are not states an agent can simply produce at will.

So the question is whether we can strategically bring about genuine states of partial uncontrol, individually or collectively. But the answer seems to be that we do this all the time. I would add to Frank's account the suggestion that we can do it without sacrificing rationality--even in the short term, to the extent that we undergo an emotion. The point depends on understanding emotions themselves as potentially rational, and as rationalizing action, but also on appreciating features of their role in directing attention that make them subject to manipulation by the agent (cf. Mele 1995).

An effective emotional strategy has to be somewhat indirect: we cannot just "talk ourselves into" sincere emotional states by commanding ourselves to feel them or by taking immediate aim at the signs of emotion. Instead, we set up the conditions under which they would arise. These include conditions of thought, as ways of understanding the situation.

Imagine working up an emotion in order to argue effectively. A case I used in Greenspan (2000) involves letting oneself get angry about a consumer complaint when the time comes to confront someone at the store. But one might also consider the everyday occasions in teaching or giving a paper when one works up a strategic emotion: indignation at opposing views or enthusiasm for a question that is not of current concern. This might be (and remain) just a pretense, but it also might be a real result of focusing attention on aspects of the material and the surrounding situation that are likely to generate the requisite emotion: the wrongheadedness of the opposing view or the features of one's own view that initially made it seem worth defending.

Some people are better at this than others, needless to say. Part of what it takes is control over attention--not entertaining those self-defeating thoughts one may well have on hand, of sympathy for the opposing view and disgust at the imperfections of one's own. Also, it is crucial not to focus too directly on strategy: the aim of generating an emotion in oneself that will enhance performance.

To count this restriction of attention as rational in decision-theoretical terms, we have to be clear that rationality does not require being immediately and centrally aware of everything one knows. On standard assumptions in decision theory it does require full knowledge pertaining to the choice situation. But putting blinders on oneself in this sense (or passively accepting them) is "within reason," as long as what one knows remains within reach, perhaps on the periphery of attention.

A perspectival evaluative account of emotions and emotional rationality can accommodate such maneuvers. We have two competing partial views of the situation--one focusing on positive features, one on negative--and some choice about where to focus. So for different purposes we can shift the focus, thereby generating the relevant emotional state--not with any certainty but with enough probability for the maneuver to count as an exercise of rational control. Since the state involves positive or negative affect (or possibly a mixture of both), we have essentially modified the payoff structure by adding emotion to the preexisting situation.

This account is compatible with the usefulness of emotions as commitment devices, which requires that our control over them be incomplete. For purposes of commitment it is enough that emotions be difficult to fake convincingly and that they make certain courses of action (or refraining from action) difficult. This means that emotional states, or those that are useful as commitment devices, have to exhibit a certain momentum: we cannot so easily just talk ourselves out of them, including those we managed to talk ourselves into.

In the case of generating enthusiasm to give a paper, momentum (with any luck) will be a product of the rewarding aspect of the feeling, as augmented by others' positive responses. But in a commitment case like the prisoner's dilemma something more is possible: if the prisoners have sworn loyalty to each other and firmed up their feelings accordingly, that means instilling negative reactions to the thought of betrayal that will be hard to go against when the opportunity to gain by confessing arises.

This case is in some ways easier in that it involves inducing dispositions to feel certain emotions rather than occurrent emotions. The notion of cultivating certain traits in ourselves, including emotional dispositions, is familiar from Aristotle's account of habituation in virtue (see 1925) and still plays a role in philosophers' views of emotional motivation (see, e.g., Wollheim 1991). However, most recent approaches to emotion focus on occurrent states, perhaps in reaction to an earlier overemphasis on dispositions in philosophical behaviorism (see Ryle 1949). To appreciate the rational role of emotion we need to consider uses of occurrent emotions such as anger (to back up a threat to call the Better Business Bureau, in my consumer case), as well as the sorts of commitments an agent can make to himself--to follow through on a resolution (possibly by "holding onto" his feelings of anger). Damasio's point about "marking" thoughts with emotion covers some cases of emotional self-commitment. Consider learning a foreign language: one way of implanting information in long-term memory is to focus emotional attention on it--perhaps by thinking of some association that evokes amusement. Examples involving practical reasoning include viewing the prospect of going back on one's resolutions--analogous to betrayal in the prisoner's dilemma--with self-disgust or some other version of anticipatory guilt, bringing on a motivating emotion now as well as generating a disposition to feel it later.

6. Emotions as Reasons

By inducing an occurrent emotion, I would say, we give ourselves a further reason for action--much as we do when we make a promise to others. We can count an emotion as itself a reason in the sense of a motivating reason, a state of thought, or as a source of reasons in a more objective sense, of normative reasons as facts about what needs to be done to sustain or ameliorate a state of emotional comfort or discomfort. However, this rational role of emotions may seem hard to accommodate within the traditional division of views within modern moral philosophy. On the one hand, though it carries echoes of Kantian notion of making law for oneself, it runs very much against the usual picture of Kant's emotionless notion of practical reason (see esp. Kant 1981: 399; but cf. Sherman 1990). On the other hand, while the contrasting Humean picture accords emotions a necessary practical role, this seems to depend on undermining the role of practical reason.

However, here as elsewhere an alternative model is provided by Aristotle (see 1925). Among other things, Aristotelian ethics suggests a picture of practical reasoning that includes not just the calculation of means to ends but also the determination of constituents of valued wholes. Aristotle had in mind happiness, of course, but one might fill in something shorter-term for particular emotions. Action expressive of an emotion has a certain value for the agent as the completion of a more integrated psychological whole: a fit between inner and outer states of oneself, as the original condition from which adult reactions developed.

For that matter, action on emotion counts as an instance of value-expressive behavior to the extent that emotions amount to felt evaluations: they have evaluative thought contents as well as amounting to affective states assessable from the standpoint of the agent. So there are two layers of evaluation that we can appeal to here. My suggestion is not that either of them is final. Expressing one's feelings is certainly not always the most rational thing to do. For one thing, the evaluations that emotions register need not represent one's considered view of the situation. But there is always pro tanto reason for expressing emotion, in the way thinking that p is a pro tanto reason for saying "p," even if it is ultimately defeated by countervailing reasons. That it manifests a state of mind is a consideration in its favor.

The reasons that emotions provide on this account are supposed to be justifying reasons, reasons in a normative sense--good reasons, at least as far as they go. There is some resistance to according emotions this role, in part because it seems to cross categories of emotion-versus-reason, but also because of its apparent implications for moral motivation. Motivation by moral emotion, if understood in the terms I have suggested, can seem suspiciously self-regarding--as if the agent's goal in acting morally were either to improve her state of comfort or discomfort or just to express herself. However, if the content of emotion is a moral evaluation or an evaluation that has moral worth (such as one that recognizes the value of another person), I think this objection can be met, at least for any reasonably broad understanding of moral motivation.

One might want to say that emotions themselves can at best just heighten the force of nonemotional reasons: whatever reasons there are for the evaluative content of emotion, or for holding it in mind. But consider how undertaking an obligation by promising or the like typically augments such reasons as already may have been in force for acting in the way that the obligation prescribes. Affect adds a further factor for an agent to take into account, a psychological factor under limited control.

Within moral philosophy attention to emotions as subjects of moral judgment has generally emphasized their role as elements of personal moral worth (see, e.g., Oakley 1992), on the assumption that their uncontrolled aspect makes them unfit subjects of moral requirement. Indeed, emotions are often invoked in explaining weakness of will. However, insofar as they involve more or less immediate comfort or discomfort, they can serve to modify the payoff structure of the situation prompting weakness of will by reversing temporal discounting (see Frank 1988, Greenspan 1988; cf. Elster 1999). Their uncontrolled aspect can itself serve as a check against weakness of the sort that involves recalculating one's reasons when the time for action arrives. Emotion provides a reason for action that is itself somewhat recalcitrant to reason, though subject to indirect control via control over attention.

The idea of emotions as factors in practical reasoning might seem to set up a dilemma. On the one hand, if their rational role is analogous to that of belief, then they are reasons only in a subjective sense, not as facts about the world (including ourselves) but as purported representations of facts. My feeling in the consumer case that the store has treated me unfairly would seem to justify threatening them with a complaint only on the assumption that it reflects the truth of the matter. On the other hand, to the extent that we think of emotions as providing new facts that we need to consider in practical reasoning--my angry feelings in the consumer case as unpleasant feelings I ought to act to assuage--they seem to play an obvious but trivial role, in the manner of "brute" facts like the fact that one has a headache, viewed as a reason (say) for staying home from a disco.

However, consider what it is like to hate the disco scene and to feel repugnance at the thought of being part of it. The evaluative and affective aspects of emotion (or the facts they represent or constitute) reinforce each other as part of a single reason for declining an invitation to a disco: it would be awful, hence I could not stand it, hence it would be all the more awful--and so on. This combined reason is not subject to either side of the alleged dilemma, and it provides a firmer barrier to persuasion than either evaluation or affect alone could hope to do.

Philosophers who accept a desire/belief account of intentional action--a contemporary reading of the practical syllogism that is often thought of as Humean--sometimes interpret the motivational role of emotions in terms of desire (see, e.g., Marks 1982, Gordon 1987, Wollheim 1991, Green 1992). Emotions of course can give rise to desires--a desire to avoid discos, in the case just outlined. In the contemporary philosopher's sense of desires as simple "wants," which leaves out the element of emotional affect, there is a trivial sense in which this would have to be true of any emotions that result in action. However, emotions do not simply reduce to desires, or desires/belief pairs; we need to bring in affect to explain their full rational bearing on action. Nor need the reasons for emotions depend on the agent's (prior) desires: there may be reasons for feeling something--a moral requirement of compassion toward others or guilt at wrong action--that apply to us regardless of our desires.

It may concern some moral philosophers that the bearing of emotions on action according to this sort of account would depend on our actually feeling them. Whether we do or not is of course a contingent matter and not subject to complete control. However, presumably the general moral requirement that is satisfied by feeling a given emotion would also require (perhaps as a second-best alternative) that one muster whatever other motivational resources one can to perform the kind of act that typically results from that emotion: aiding those in distress, say, for compassion, or making up for past wrongs, in the case of guilt.

It is important not to confuse the moral or other practical reason to feel a given emotion--whether as a reason for action or just in itself, as part of a personal ideal of virtue--with the reasons why it is appropriate to feel. On the perspectival account outlined here the latter are analogous to evidence for rational belief, except that we do not assume it is irrational not to feel an appropriate emotion. At this stage, if rational warrant is all that is in question, we have something like a theoretical syllogism yielding sufficient but noncompelling reason for an emotion. Its conclusion serves, or can serve, as the first premise of a practical syllogism leading to action.

7. Evaluative Affects

My use of the propositional attitudes approach to emotions has been directed in the first instance toward questions of rationality. This is usually taken as a later question, but I think that an answer to it can be read back onto the philosopher's traditional first question, about the nature of emotions, for an answer that does not pull quite so sharply away as judgmentalism did from the traditional approach to emotions in terms of feelings.

For recent attempts to reestablish the feeling theory see Leighton (1985), Robinson (1995), and Pugmire (1998). However, emotions can be feelings and still be propositional attitudes. The propositional attitudes approach need not be understood as assigning to emotions a separable element of propositional thought as distinct from feeling. My suggestion is rather that emotional affect itself renders a judgment that can be stated in propositional terms--by us theorists, that is, rather than necessarily by the agent.

The view depends on assigning the affective element of emotion an intentional content. The assumption of intentionality at this level of basic feeling can sound mysterious, but in principle it is no more so than in familiar cases involving units of language and thought. Insofar as emotional comfort and discomfort does have a content, moreover, the affective element of emotion should not be equated with simple hedonic states of pleasure or pain. Instead of thinking of "content" on the metaphor of containment in a composite entity, the idea of emotional affect as pointing toward something outside itself (though internal to the composite entity, of affect plus evaluation, that we think of as the emotion) might be less misleading, with bodily gestures as the appropriate analogue.

The appropriate cognitive analogue, rather than linguistic meaning, would seem to be reference--as in the thesis from Brentano 1973 that made intentionality an issue, the "intentional reference of the mental." We need not assume that the relevant referential relationship is type-to-type--that a given sort of affect always refers to the same proposition--as cases of demonstrative or other indexical reference make clear.

For emotions like love and hatred whose surface structure is not propositional (one need not love the fact that the love object has such-and-such a feature), we can distinguish propositional components of emotion corresponding to its motivational role: discomfort that one is far from the love object, say, corresponding to the desire for closeness characteristic of love (Greenspan 1988). My own rather minimal assumption has been that the positive or negative aspect of a given feeling--understood in terms of motivational significance rather than hedonic tone--represents the positive or negative aspect of the corresponding evaluation. However, there are ways of further specifying this simple bipolar account, which was set up just for purposes of getting at the rational role of emotions.

The evidence cited by Griffiths (1997) from Ekman (1992)--and ultimately from Darwin (see 1965)--groups our basic emotions into multiple "affect programs" that evolved initially to prepare the body for different modes of action: fight or flight, for instance, in the familiar cases of anger and fear. The element of feeling in these clusters of responses come to be signs of, and in that sense "about," the need for a certain kind of action. This natural referent of feeling would then be subject to social modification, including the sort of individual interaction with caretakers that de Sousa's account of paradigm scenarios takes as explaining the intentionality of full-fledged emotions.

This account might be extended to cover purely expressive action, without evolutionary function, possibly mediated by symbols based on social learning. A more mentalistic variant might include or substitute reference to an associated mental act such as directing attention--or simply the act of taking some considerations as reasons for action (cf. Scanlon forthcoming). At any rate, given the broad use of "judgment" in the emotions literature, the general suggestion that feeling itself can have an evaluative intentional content may be thought of as a friendly amendment to judgmentalism. There are parallels to it in the continental literature (cf. Brentano 1973: 199; cf. 223), though with emotions presented in contrast with judgments. In general, the view that "affect evaluates" (as I would sum it up in slogan form) affords a way of accommodating empirical findings (as in Zajonc 1980, LeDoux 1996) that some cases of emotion such as primitive fear responses do not involve cognition (cf. Deigh 1994, Robinson 1995). Cases of "gut feelings" recording good reasons for action suggest that even inarticulate emotions sometimes act in aid of rationality.


Aristotle. 1925. The Nicomachean Ethics. Oxford: Oxford University Press.

___. 1941. Rhetorica. In R. McKeon, ed. The Basic Works of Aristotle. New York: Random House.

Averill, J. 1980. "A Constructivist View of Emotion." In R. Plutchik and H. Kellerman, eds. Emotion: Theory, Research and Experience, vol. 1: Theories of Emotion. New York: Academic Press.

Bedford, E. 1957. "The Emotions." Proceedings of the Aristotelian Society 57: 281-304.

Ben-Ze'ev, A. 2000. The Subtlety of Emotions. Cambridge, Mass.: M. I. T. Press.

Blackburn, S. 1998. Ruling Passions: A Theory of Practical Reasoning. Oxford: Clarendon Press.

Brentano, F. 1973. Psychology from an Empirical Standpoint. London: Routledge and Kegan Paul.

Budd, M. 1985. Music and the Emotions. London: Routledge & Kegan Paul.

Cosmides, L. and Tooby, J. 1992. "Cognitive Adaptations for Social Exchange." In J. Barkow, L. Cosmides, and J. Tooby, eds. The Adapted Mind: Evolutionary Psychology and the Generation of Culture. Oxford: Oxford University Press.

Damasio, A. 1994. Descartes' Error: Emotion, Reason, and the Human Brain. New York: Putnam.

D'Arms, J. and Jacobson, D. 2000. "Sentiment and Value." Ethics 110: 722-48.

Darwin, C. 1965. The Expression of the Emotions in Man and Animals. Chicago: University of Chicago Press.

Davidson, D. 1980. "Hume's Cognitive Theory of Pride." In Essays on Actions and Events. Oxford: Oxford University Press.

Davis, W. 1987. "The Varieties of Fear." Philosophical Studies. 51: 287-310.

Deigh, J. 1994. "Cognitivism in the Theory of Emotions." Ethics 104: 824-54.

de Sousa, R. 1987. The Rationality of Emotion. Cambridge, Mass.: M.I.T. Press.

Descartes, R. 1970. The Passions of the Soul. In The Philosophical Works of Descartes, Vol. I. Cambridge: Cambridge University Press.

Ekman, P. 1992. "Are There Basic Emotions?" Psychological Review 99:550-53.

Elster, J. 1999. Alchemies of the Mind: Rationality and the Emotions. Cambridge: Cambridge University Press.

Frank, R. 1988. Passions Within Reason: The Strategic Role of the Emotions. New York: Norton.

Gordon, R. 1987. The Structure of Emotions: Investigations in Cognitive Philosophy. Cambridge: Cambridge University Press.

Gibbard, A. 1990. Wise Choices, Apt Feelings: A Theory of Normative Judgment. Cambridge, Mass.: Harvard University Press.

Green, O. 1992. The Emotions: A Philosophical Theory. Dordrecht: Kluwer Academic Publishers.

Greenspan, P. 1980. "A Case of Mixed Feelings: Ambivalence and the Logic of Emotion." In A. Rorty, ed. Explaining Emotions. Berkeley: University of California Press.

___. 1981. "Emotions as Evaluations." Pacific Philosophical Quarterly, 62: 158-69.

___. 1988. Emotions and Reasons: An Inquiry into Emotional Justification. New York: Routledge, Chapman and Hall.

___. 1995. Practical Guilt: Moral Dilemmas, Emotions, and Social Norms. New York: Oxford University Press.

___. 1998. "Moral Responses and Moral Theory: Socially Based Externalist Ethics." Journal of Ethics 2: 103-22.

___. 2000. "Emotional Strategies and Rationality." Ethics 3: 469-87.

Griffiths, P. 1997. What Emotions Really Are: The Problem of Psychological Categories. Chicago: University of Chicago Press.

Harris, P. 1989. Children and Emotion: The Development of Psychological Understanding. Oxford: Basil Blackwell.

Hume, D. 1978. A Treatise of Human Nature. Oxford: Clarendon Press.

James, W. 1884. "What Is an Emotion?" Mind 9: 188-205.

Kant, E. 1981. Grounding for the Metaphysics of Morals. Indianapolis: Hackett.

Kenny, A. 1963. Action, Emotion and Will. New York: Humanities.

LeDoux, J. 1996. The Emotional Brain: The Mysterious Underpinnings of Emotional Life. New York: Simon & Schuster.

Leighton, S. 1985. "Feelings and Emotion." Review of Metaphysics 38: 303-21.

Lutz, C. 1986. "The Domains of Emotion Words on Ifaluk." In Harré, ed. 1986. The Social Construction of the Emotions. London: Oxford University Press.

Lyons, W. 1980. Emotion. Cambridge: Cambridge University Press.

Marks, J. 1982. "A Theory of Emotions." Philosophical Studies, 42 227-42.

Mele, A. 1995. Autonomous Agents. New York: Oxford University Press.

Murdoch, I. 1970. The Sovereignty of Good. London: Routledge & Kegan Paul.

Neu, J. 1977. Emotion, Thought and Therapy. Berkeley: University of California Press.

Nussbaum, M. 1993. The Therapy of Desire. New York: Cambridge University Press.

Pitcher, G. 1965. "Emotion." Mind 74: 326-45.

Pugmire, D. 1998. Rediscovering Emotion. Edinburgh: Edinburgh University Press.

Roberts, R. 1988. "What an Emotion Is: A Sketch." Philosophical Review 79: 183-209.

Robinson, R. 1995. "Startle." Journal of Philosophy 92: 53-74.

Rorty, A. 1980. "Introduction." In A. Rorty, ed., Explaining Emotions. Berkeley: University of California Press.

Ryle, G. 1949. The Concept of Mind London: Hutchinson.

Scanlon, T. forthcoming. "Passions and Reasons." In Buss, S., and Overton, L., eds. The Contours of Agency. Cambridge, Mass.: M. I. T. Press.

Sartre, J. 1962. Sketch for a Theory of the Emotions New York: Methuen.

Sherman, N. 1990. "The Place of Emotions in Kantian Morality." In O. Flanagan and A. Rorty, eds., Identity, Character, and Morality. Cambridge, Mass.: M. I. T. Press.

Solomon, R. 1973. "Emotions and Choice." Review of Metaphysics 27: 20-41.

___. 1976. The Passions. Garden City, New York: Doubleday.

Stocker, M. 1987. "Emotional Thoughts." American Philosophical Quarterly 24: 59-69.

Stocker, M. 1996. Valuing Emotions. Cambridge: Cambridge University Press.

Wollhaim, R. 1991. On the Emotions. New Haven: Yale University Press.

Zajonc, R. 1984. "On the Primacy of Affect." American Psychologist 39: 117-23.

1. For comments on earlier versions of this paper, I owe thanks to Erich Diese, Scott James, Stephen Leighton, and Jerrold Levinson.